The zebrafish offers unique advantages as a vertebrate system to analyze cell and tissue structure and development in vivo. External fertilization and transparent larvae allow imaging of cells in live, developing animals. New genetic techniques can knock out most zebrafish genes, making it critical to develop high- throughput techniques to analyze mutant phenotypes (the phenome project). For all these experiments, confocal microscopy is an essential tool, making it equally critical to develop software to analyze 3D and 4D (3D over time) confocal data and quickly derive biologically-relevant conclusions, both qualitative and quantitative. Image analysis requires five steps: preprocessing, registration, visualization, segmentation, and quantitation. We have already developed a downloadable confocal analysis application, FluoRender, and optimized it for visualization, providing significant advantages over existing commercial applications. The present proposal would make FluoRender a full-featured package by adding the other four steps. Our goal is to enable the biologist to easily analyze 3D and 4D confocal data, identifying and measuring features of interest, and allowing rapid repetitive analysis of multiple samples. For features like segmentation that benefit from automation, we provide user validation and editing, emphasizing high accuracy rather than pure speed. We will focus on three specific problems: 1) 3D image mosaicking, 4D drift removal; 2) 3D segmentation of confocal data; and 3) 4D tracking in confocal data. Mosaicking allows scanning of specimens larger than a microscope's field of view. Timelapse experiments benefit from 4D drift removal, since the growing embryo can change shape or the microscope's focus can drift. We will develop segmentation methods for objects from three classes: nuclei/cells, axons/dendrites, and tissues. We will develop semi-automatic segmentation methods for zebrafish labeled with multiple fluorophores, or spectrally (Brainbow). We will develop methods so that once cells or tissues are initially segmented, they can be tracked over time, then visualized co-registered with the raw data, providing context for interpreting their motion. The improved FluoRender software will provide a freely-distributed, portable suite that enables rapid analysis of 3D or 4D confocal datasets. This will aid in analyzing cell movements, neuronal circuitry, and tissue development in wildtype and mutant embryos. Given the growing relevance of zebrafish as a human disease model, the proposed analysis software can be expected to benefit our understanding of many different developmental, neurobiological, and metabolic diseases.